This dataset supports an analysis of over 29,000 California local elections from 1995 to 2021 where ballot order was randomly assigned. It examines the causal effect of being listed first on vote share, with a focus on differences by candidate gender and ethnicity. The analysis finds a first-listing premium that varies for non-white candidates based on electorate composition and election timing.
Use Cases
- Estimate the causal effect of randomly assigned ballot order on candidate vote share using over 29,000 election records.
- Analyze how the vote share premium from being listed first differs by candidate gender and ethnicity.
- Investigate how the advantage for non-white candidates varies with the partisan and racial composition of the electorate.
- Model the interaction between election timing and the ballot order effect on marginalized candidate representation.
Strengths
- Leverages a natural experiment from over 29,000 California local elections for causal inference.
- Covers a 26-year time range from 1995 to 2021, allowing for longitudinal analysis.
- Focuses on specific, policy-relevant variables: ballot order, candidate gender, and candidate ethnicity.
Limitations
- The specific columns, sample data, and file formats are unknown, limiting immediate usability.
- Geographic scope is limited to California, which may not generalize to other political contexts.
- The dataset's structure and required join keys for replication are not described.
Provenance
- Source
- Political Behavior Dataverse
- Collection Method
- Leverages a natural experiment from elections where ballot order was randomly assigned.
- Time Range
- 1995 to 2021
- Freshness
- null
- Geography
- California, United States